[HW Accel Support]: Intel GPU always <5% , CPU relatively high load ~70% + Fine tuning a 24 camera set up #11894
Answered
by
NickM-27
Bashewa
asked this question in
Hardware Acceleration Support
-
Describe the problem you are having24 mixed Camera system is running well enough, but wondering, if it possible for the Intel GPU to take over some of the CPU load ? Running as HAS addon Intel NUC NUC7I3BNH I3 7100U 8GB with 4TB western digital Hard drive Version0.13.2-6476F8A Frigate config filemqtt:
enabled: True
host: xx
user: xx
password: xx
ffmpeg:
hwaccel_args: preset-vaapi
# record: preset-record-generic-audio-copy
detectors:
coral1:
type: edgetpu
device: usb:0
coral2:
type: edgetpu
device: usb:1
record:
#sync_recordings: True ##--- resync DB if suspected files have been deleted that frigate doesnt know about, CPU INTENSIVE!! DONT LEAVE ON ##
enabled: True
retain:
days: 7
mode: motion
events:
retain:
default: 7
mode: active_objects
objects:
dog: 2
car: 7
cat: 2
person: 7
objects:
track:
- person
- dog
- cat
- car
snapshots:
enabled: True
retain:
default: 7
## default universal below, set at each camera to match ffprobe substream
detect:
enabled: true
width: 640
height: 480
fps: 5
# # Set the birdseye view to always show a live stream of the cameras
# birdseye:
# mode: continuous
birdseye:
enabled: True
mode: motion
## stream=0.sdp? is the HD stream and stream=1.sdp? is the SD stream. for the ICSEE cams
############
go2rtc:
streams:
Garage_Freezers:
- rtsp://admin:[email protected]:554/onvif1 # <- onvif 1 --HD ,,, onvif 2 lower quality
# Garage_Freezers_sub: ### 320 x 180 substream to small
# - rtsp://admin:[email protected]:554/onvif2
House_Gate:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
House_Gate_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
House_Garage:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
House_Garage_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
Office_Gate:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
Office_Gate_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
Lab_Door:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
Lab_Door_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
Middle_Garage:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
Middle_Garage_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
Factory_Outside:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
Factory_Outside_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
KC_Store:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
KC_Store_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
Thatch_Office:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
Thatch_Office_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
House_Garage_Inside:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
House_Garage_Inside_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
Inverter_Cam2: ## this ptz the cam that cant swivel and head falls off
- rtsp://admin:[email protected]:554/onvif1 #
# Inverter_Cam3: #Wonky cam , needs the bin
# - rtsp://admin:[email protected]:554/onvif1
Bottom_Dam:
- rtsp://admin:[email protected]:554/onvif1 #
# Bottom_Dam_sub: ### 320 x 180 substream , to small for detection
# - rtsp://admin:[email protected]:554/onvif2
SriCam_PTZ:
rtsp://admin:[email protected]:554/onvif1
# SriCam_PTZ_sub: #### substream to small so not using 320 x 180
# rtsp://admin:[email protected]:554/onvif2
Bottom_cam_1:
rtsp://admin:[email protected]:554/onvif1
Inverter_Sonoff:
rtsp://rtsp:[email protected]:554/av_stream/ch0
Inverter_Sonoff_sub:
rtsp://rtsp:[email protected]:554/av_stream/ch1
Bottom_cam_3:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=0.sdp?
Bottom_cam_3_sub:
- rtsp://admin:[email protected]:554/user=admin&password=password&channel=1&stream=1.sdp?
######
###### set hikvision streams to channel--02 SD as already recording all HD stream on HiKvision NVR--channel 01 is HD
Factory_1:
- rtsp://admin:[email protected]/ISAPI/Streaming/channels/102
Factory_2:
- rtsp://admin:[email protected]/ISAPI/Streaming/channels/202
Factory_3:
- rtsp://admin:[email protected]/ISAPI/Streaming/channels/302
Factory_4:
- rtsp://admin:[email protected]/ISAPI/Streaming/channels/402
Factory_5:
- rtsp://admin:[email protected]/ISAPI/Streaming/channels/502
Factory_6:
- rtsp://admin:[email protected]/ISAPI/Streaming/channels/602
Factory_7:
- rtsp://admin:[email protected]/ISAPI/Streaming/channels/702
Factory_8:
- rtsp://admin:[email protected]/ISAPI/Streaming/channels/802
cameras:
#### HikVision NVR
Factory_1:
enabled: true
detect:
width: 352
height: 288
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_1
input_args: preset-rtsp-restream
roles:
- record
- detect
motion:
mask:
- 179,0,184,53,0,55,0,0 ## HIKVision Time Stamp
#### HikVision NVR
Factory_2:
enabled: true
detect:
width: 352
height: 288
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_2
input_args: preset-rtsp-restream
roles:
- record
- detect
motion:
mask:
- 179,0,184,53,0,55,0,0 ## HIKVision Time Stamp
#### HikVision NVR
Factory_3:
enabled: true
detect:
width: 352
height: 288
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_3
input_args: preset-rtsp-restream
roles:
# - motion
- detect
motion:
mask:
- 179,0,184,53,0,55,0,0 ## HIKVision Time Stamp
#### HikVision NVR
Factory_4:
enabled: true
detect:
width: 352
height: 288
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_4
input_args: preset-rtsp-restream
roles:
# - motion
- detect
motion:
mask:
- 179,0,184,53,0,55,0,0 ## HIKVision Time Stamp
#### HikVision NVR
Factory_5:
enabled: true
detect:
width: 352
height: 288
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_5
input_args: preset-rtsp-restream
roles:
# - motion
- detect
motion:
mask:
- 179,0,184,53,0,55,0,0 ## HIKVision Time Stamp
#### HikVision NVR
Factory_6:
enabled: true
detect:
width: 352
height: 288
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_6
input_args: preset-rtsp-restream
roles:
# - motion
- detect
motion:
mask:
- 179,0,184,53,0,55,0,0 ## HIKVision Time Stamp
#### HikVision NVR
Factory_7:
enabled: true
detect:
width: 352
height: 288
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_7
input_args: preset-rtsp-restream
roles:
# - motion
- detect
motion:
mask:
- 179,0,184,53,0,55,0,0 ## HIKVision Time Stamp
#### HikVision NVR
Factory_8:
enabled: true
detect:
width: 352
height: 288
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_8
input_args: preset-rtsp-restream
roles:
- record
- detect
objects:
track: ##### Stop detecting parked cars
- person
- dog
- bicycle
- cat
- mouse
motion:
mask:
- 179,0,184,53,0,55,0,0 ## HIKVision Time Stamp
###
Inverter_Sonoff: # <--- this will be changed to your actual camera later
enabled: true
detect:
width: 640
height: 380
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Inverter_Sonoff
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/Inverter_Sonoff_sub
input_args: preset-rtsp-restream
roles:
- detect
motion:
mask:
- 0,21,160,22,158,0,0,0 ###mask time stamp
### YYPTP PTZ CAM
Garage_Freezers:
enabled: true
detect:
width: 1280
height: 720
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Garage_Freezers
input_args: preset-rtsp-restream
roles:
- record
# - path: rtsp://127.0.0.1:8554/Garage_Freezers_sub
# input_args: preset-rtsp-restream
# roles:
- detect
motion:
mask:
- 496,0,493,43,0,41,0,0 ##mask time stamp
#### ICSEE Non PTZ
House_Gate:
enabled: true
detect:
width: 704
height: 576
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/House_Gate
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/House_Gate_sub
input_args: preset-rtsp-restream
roles:
- detect
motion:
mask:
- 640,0,640,60,401,62,398,0 ## Mask time stamp#### ICSEE Non PTZ
House_Garage:
enabled: true
detect:
width: 704
height: 576
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/House_Garage
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/House_Garage_sub
input_args: preset-rtsp-restream
roles:
- detect
motion:
mask:
- 640,0,640,60,401,62,398,0 ## Mask time stamp
zones:
top_driveway_house_cam:
coordinates: 433,0,431,116,241,112,290,0
objects:
- person
filters:
person:
# min_area: 4000
# max_area: 100000
# min_ratio: 0.5
# max_ratio: 1.0 ## ignore objects wider than they tall
# min_score: 0.5
threshold: 0.6
house_garage_cam_all: ## just under time stamp
coordinates: 640,56,640,480,0,480,0,63
objects:
- person
- cat
- dog
- car
filters:
person:
min_area: 20000
# max_area: 100000
# min_ratio: 0.5
max_ratio: 1.0 ## must be taller than wide
# min_score: 0.5
threshold: 0.7
cat:
min_area: 400
# max_area: 100000
# min_ratio: 0.5
# max_ratio: 2.0
# min_score: 0.5
threshold: 0.7
dog:
min_area: 400
# max_area: 100000
# min_ratio: 0.5
# max_ratio: 2.0
# min_score: 0.5
threshold: 0.7
#### ICSEE Non PTZ
Office_Gate:
enabled: true
detect:
width: 704
height: 576
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Office_Gate
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/Office_Gate_sub
input_args: preset-rtsp-restream
roles:
- detect
motion:
mask:
- 640,0,640,60,401,62,398,0 ## Mask time stamp
zones:
office_gate_motor: ### track ppeople around gate motor
coordinates: 403,296,419,480,247,480,254,301
inertia: 7 ## person must be in frame for 1.5 seconds
# Optional: List of objects that can trigger this zone (default: all tracked objects)
objects:
- person
front_of_office_gate:
coordinates: 398,109,543,344,396,386,280,86
objects:
- person
cars_infront_of_office_gate:
coordinates: 398,109,543,344,396,386,280,86
inertia: 20 ## car must be in front of gate for 4 seconds
objects:
- car
objects: #### add non zone filters last
filters:
person:
mask:
- 640,123,640,232,350,89,342,0 ### mask people in the road
cat:
mask:
- 618,46,608,94,544,94,539,42 ### add what ever mask for objects iones tracking -- arb piece of veldt here
#### ICSEE Non PTZ
Lab_Door:
enabled: true
detect:
width: 640
height: 360
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Lab_Door
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/Lab_Door_sub
input_args: preset-rtsp-restream
roles:
- detect
motion:
mask:
- 640,0,640,82,0,90,0,0 ## mask roofs
#### ICSEE Non PTZ
Middle_Garage:
enabled: true
detect:
width: 704
height: 576
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Middle_Garage
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/Middle_Garage_sub
input_args: preset-rtsp-restream
roles:
- detect
motion:
mask:
- 640,0,640,60,401,62,398,0 ## Mask time stamp
#### ICSEE Non PTZ
Factory_Outside:
enabled: true
detect:
width: 704
height: 576
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Factory_Outside
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/Factory_Outside_sub
input_args: preset-rtsp-restream
roles:
- detect
motion:
mask:
- 640,0,640,60,401,62,398,0 ## Mask time stamp
#### ICSEE Non PTZ
KC_Store:
enabled: true
detect:
width: 640
height: 360
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/KC_Store
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/KC_Store_sub
input_args: preset-rtsp-restream
roles:
- detect
motion: ## mask out time stamp
threshold: 2 #### still getting fake bags , but under 80% mmmmmm why , so can delete line again
mask:
- 640,0,640,65,390,63,392,0
- 640,0,640,454,361,480,539,0 ### bags on the side
- 0,480,0,63,332,0,167,480 ### bags on the side
zones:
kc_store_entire_view:
coordinates: 640,96,640,480,0,480,0,106
filters:
person: ## give person threshhold higher than bags it picks up
# min_area: 5000
# max_area: 100000
threshold: 0.8
#### ICSEE Non PTZ
Thatch_Office:
enabled: true
detect:
width: 704
height: 576
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Thatch_Office
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/Thatch_Office_sub
input_args: preset-rtsp-restream
roles:
- detect
motion: ## mask out time stamp
mask:
- 640,0,640,65,390,63,392,0
#### ICSEE Non PTZ
House_Garage_Inside:
enabled: true
detect:
width: 704
height: 576
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/House_Garage_Inside
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/House_Garage_Inside_sub
input_args: preset-rtsp-restream
roles:
- detect
objects:
track: ##### Stop detecting at parked cars
- person
- dog
- bicycle
- cat
- mouse
motion: ## mask out time stamp
mask:
- 640,0,640,65,390,63,392,0
#### ICSEE Non PTZ
Inverter_Cam2: # Wonky Cam , to bin or not to bin, that is the question
enabled: true
detect: ### not using sub stream on these
width: 1280
height: 720
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Inverter_Cam2
input_args: preset-rtsp-restream
roles:
- record
- detect
motion:
mask:
- 0,0,245,0,245,27,0,29 ## time stamp
#### ICSEE Non PTZ
# Inverter_Cam3: ## check out whether salvagable or throw away
# enabled: true
# ffmpeg:
# inputs:
# - path: rtsp://127.0.0.1:8554/Inverter_Cam3
# input_args: preset-rtsp-restream
# roles:
# - record
# - detect
#### ICSEE Non PTZ
Bottom_Dam: ##### Low Res feed so not bothering to split record and detect
enabled: true
detect: ### not using sub stream on these
width: 1280
height: 720
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Bottom_Dam
input_args: preset-rtsp-restream
roles:
- record
- detect
motion:
mask:
- 640,0,640,107,0,104,0,0 ## Skyline and Time stamp
zones:
bottom_dam_pump_and_db:
coordinates: 284,177,329,354,92,344,96,187
# Optional: Number of consecutive frames required for object to be considered present in the zone (default: shown below).
inertia: 3
# Optional: List of objects that can trigger this zone (default: all tracked objects)
objects:
- person
# Optional: Zone level object filters.
# NOTE: The global and camera filters are applied upstream.
filters:
person:
min_area: 18000
# max_area: 100000
# min_ratio: 0.5
max_ratio: 1.0 ## if max_ratio was set to 1 then any object that is wider than it is tall would be ignored.
# min_score: 0.5
threshold: 0.50 ## setting low as going to set max ratio
#### SriCam PTZ
SriCam_PTZ:
enabled: true
detect:
width: 640
height: 360
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/SriCam_PTZ
input_args: preset-rtsp-restream
roles:
- record
### Sricam substream is smaller than 320 x 320 so use main stream
# - path: rtsp://127.0.0.1:8554/SriCam_PTZ_sub
# input_args: preset-rtsp-restream
# roles:
- detect
motion: ## mask out time stamp
mask:
- 320,0,316,67,0,77,0,0
#### ICSEE Non PTZ
Bottom_cam_1: ##### Low Res feed so not bothering to split record and detect
enabled: true
detect: ### not using sub stream on these
width: 1280
height: 720
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Bottom_cam_1
input_args: preset-rtsp-restream
roles:
- record
- detect
motion:
mask:
- 487,0,485,53,0,50,0,0 ## Skyline and Time stamp
#### ICSEE Non PTZ
# Bottom_cam_2: ##### Low Res feed so not bothering to split record and detect
# enabled: true
# detect: ### not using sub stream on these
# width: 1280
# height: 720
# fps: 5
# ffmpeg:
# inputs:
# - path: rtsp://127.0.0.1:8554/Bottom_cam_2
# input_args: preset-rtsp-restream
# roles:
# - record
# - detect
# motion:
# mask:
# - 640,0,640,60,0,67,0,0 ## Mask Time Stamp and Skyline
#### ICSEE Non PTZ 180 Degree camera
Bottom_cam_3:
enabled: true
detect:
width: 1280
height: 360
fps: 5
ffmpeg:
inputs:
- path: rtsp://127.0.0.1:8554/Bottom_cam_3
input_args: preset-rtsp-restream
roles:
- record
- path: rtsp://127.0.0.1:8554/Bottom_cam_3_sub
input_args: preset-rtsp-restream
roles:
- detect
motion: ## mask out time stamp
mask:
- 1280,0,1280,45,1087,44,1087,0docker-compose file or Docker CLI command2024-06-11 20:10:43.912821917 [INFO] Preparing Frigate...
2024-06-11 20:10:44.703219675 [INFO] Starting Frigate...
2024-06-11 20:11:04.802474755 [2024-06-11 22:11:04] frigate.app INFO : Starting Frigate (0.13.2-6476f8a)
2024-06-11 20:11:04.802729016 [2024-06-11 22:11:04] frigate.app INFO : Creating directory: /tmp/cache
2024-06-11 20:11:05.117510739 [2024-06-11 22:11:05] peewee_migrate.logs INFO : Starting migrations
2024-06-11 20:11:05.169369907 [2024-06-11 22:11:05] peewee_migrate.logs INFO : There is nothing to migrate
2024-06-11 20:11:05.199210093 [2024-06-11 22:11:05] frigate.app INFO : Recording process started: 336
2024-06-11 20:11:05.206864149 [2024-06-11 22:11:05] frigate.app INFO : go2rtc process pid: 89
2024-06-11 20:11:05.368276262 [2024-06-11 22:11:05] detector.coral1 INFO : Starting detection process: 346
2024-06-11 20:11:08.587340625 [2024-06-11 22:11:05] detector.coral2 INFO : Starting detection process: 348
2024-06-11 20:11:08.599686052 [2024-06-11 22:11:05] frigate.app INFO : Output process started: 355
2024-06-11 20:11:08.600183454 [2024-06-11 22:11:05] frigate.detectors.plugins.edgetpu_tfl INFO : Attempting to load TPU as usb:0
2024-06-11 20:11:08.601225104 [2024-06-11 22:11:05] frigate.detectors.plugins.edgetpu_tfl INFO : Attempting to load TPU as usb:1
2024-06-11 20:11:08.601690298 [2024-06-11 22:11:07] frigate.detectors.plugins.edgetpu_tfl INFO : TPU found
2024-06-11 20:11:08.602079197 [2024-06-11 22:11:08] frigate.detectors.plugins.edgetpu_tfl INFO : TPU found
2024-06-11 20:11:11.958428020 [2024-06-11 22:11:11] frigate.app INFO : Camera processor started for Factory_1: 457
2024-06-11 20:11:11.965336223 [2024-06-11 22:11:11] frigate.app INFO : Camera processor started for Factory_2: 458
2024-06-11 20:11:11.966251814 [2024-06-11 22:11:11] frigate.app INFO : Camera processor started for Factory_3: 460
2024-06-11 20:11:11.973320032 [2024-06-11 22:11:11] frigate.app INFO : Camera processor started for Factory_4: 462
2024-06-11 20:11:11.982389366 [2024-06-11 22:11:11] frigate.app INFO : Camera processor started for Factory_5: 464
2024-06-11 20:11:11.996591851 [2024-06-11 22:11:11] frigate.app INFO : Camera processor started for Factory_6: 466
2024-06-11 20:11:11.996717374 [2024-06-11 22:11:11] frigate.app INFO : Camera processor started for Factory_7: 468
2024-06-11 20:11:12.008951390 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Factory_8: 471
2024-06-11 20:11:12.015770909 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Inverter_Sonoff: 472
2024-06-11 20:11:12.022580338 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Garage_Freezers: 474
2024-06-11 20:11:12.040122346 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for House_Gate: 476
2024-06-11 20:11:12.040128701 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for House_Garage: 478
2024-06-11 20:11:12.055932269 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Office_Gate: 480
2024-06-11 20:11:12.056990541 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Lab_Door: 482
2024-06-11 20:11:12.065687235 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Middle_Garage: 484
2024-06-11 20:11:12.072340370 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Factory_Outside: 486
2024-06-11 20:11:12.089594784 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for KC_Store: 488
2024-06-11 20:11:12.089600608 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Thatch_Office: 490
2024-06-11 20:11:12.102655949 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for House_Garage_Inside: 492
2024-06-11 20:11:12.127473148 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Inverter_Cam2: 494
2024-06-11 20:11:12.127612558 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Bottom_Dam: 496
2024-06-11 20:11:12.128748826 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for SriCam_PTZ: 498
2024-06-11 20:11:12.137218283 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Bottom_cam_1: 500
2024-06-11 20:11:12.149167630 [2024-06-11 22:11:12] frigate.app INFO : Camera processor started for Bottom_cam_3: 502
2024-06-11 20:11:12.162703330 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_1: 504
2024-06-11 20:11:12.163826575 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_2: 508
2024-06-11 20:11:12.191191881 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_3: 511
2024-06-11 20:11:12.191197792 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_4: 515
2024-06-11 20:11:12.194159690 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_5: 518
2024-06-11 20:11:12.205195944 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_6: 524
2024-06-11 20:11:12.246287347 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_7: 527
2024-06-11 20:11:12.246793649 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_8: 532
2024-06-11 20:11:12.255558834 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Inverter_Sonoff: 536
2024-06-11 20:11:12.295037553 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Garage_Freezers: 546
2024-06-11 20:11:12.308541226 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for House_Gate: 549
2024-06-11 20:11:12.310274282 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for House_Garage: 561
2024-06-11 20:11:12.334169786 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Office_Gate: 565
2024-06-11 20:11:12.340167200 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Lab_Door: 567
2024-06-11 20:11:12.360930203 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Middle_Garage: 577
2024-06-11 20:11:12.397071402 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Factory_Outside: 587
2024-06-11 20:11:12.397593259 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for KC_Store: 590
2024-06-11 20:11:12.422292212 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Thatch_Office: 595
2024-06-11 20:11:12.451010902 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for House_Garage_Inside: 599
2024-06-11 20:11:12.455160753 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Inverter_Cam2: 611
2024-06-11 20:11:12.474943545 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Bottom_Dam: 619
2024-06-11 20:11:12.515517035 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for SriCam_PTZ: 623
2024-06-11 20:11:12.520775651 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Bottom_cam_1: 632
2024-06-11 20:11:12.544516401 [2024-06-11 22:11:12] frigate.app INFO : Capture process started for Bottom_cam_3: 638Relevant log outputVAINFO OUTPUT
Return Code: 0
Process Output:
vainfo: VA-API version: 1.17 (libva 2.10.0)
vainfo: Driver version: Intel iHD driver for Intel(R) Gen Graphics - 23.1.1 ()
vainfo: Supported profile and entrypoints
VAProfileNone : VAEntrypointVideoProc
VAProfileNone : VAEntrypointStats
VAProfileMPEG2Simple : VAEntrypointVLD
VAProfileMPEG2Simple : VAEntrypointEncSlice
VAProfileMPEG2Main : VAEntrypointVLD
VAProfileMPEG2Main : VAEntrypointEncSlice
VAProfileH264Main : VAEntrypointVLD
VAProfileH264Main : VAEntrypointEncSlice
VAProfileH264Main : VAEntrypointFEI
VAProfileH264Main : VAEntrypointEncSliceLP
VAProfileH264High : VAEntrypointVLD
VAProfileH264High : VAEntrypointEncSlice
VAProfileH264High : VAEntrypointFEI
VAProfileH264High : VAEntrypointEncSliceLP
VAProfileVC1Simple : VAEntrypointVLD
VAProfileVC1Main : VAEntrypointVLD
VAProfileVC1Advanced : VAEntrypointVLD
VAProfileJPEGBaseline : VAEntrypointVLD
VAProfileJPEGBaseline : VAEntrypointEncPicture
VAProfileH264ConstrainedBaseline: VAEntrypointVLD
VAProfileH264ConstrainedBaseline: VAEntrypointEncSlice
VAProfileH264ConstrainedBaseline: VAEntrypointFEI
VAProfileH264ConstrainedBaseline: VAEntrypointEncSliceLP
VAProfileVP8Version0_3 : VAEntrypointVLD
VAProfileVP8Version0_3 : VAEntrypointEncSlice
VAProfileHEVCMain : VAEntrypointVLD
VAProfileHEVCMain : VAEntrypointEncSlice
VAProfileHEVCMain : VAEntrypointFEI
VAProfileHEVCMain10 : VAEntrypointVLD
VAProfileHEVCMain10 : VAEntrypointEncSlice
VAProfileVP9Profile0 : VAEntrypointVLD
VAProfileVP9Profile2 : VAEntrypointVLD
COPYFFprobe output from your cameraAll set to sub stream where applicable, if sub stream is less than 320 x 320 , main stream is used
13 x "decentish" Cameras with
FFPROBE OUTPUT
Stream 0:
Return Code: 0
Video:
Codec: H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
Resolution: 2592x1944
FPS: 12/1
Stream 1:
Return Code: 0
Video:
Codec: H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
Resolution: 704x576
FPS: 12/1
5 worst cameras
Stream 0:
Return Code: 0
Video:
Codec: H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10
Resolution: 1280x720
FPS: Unknown
Don't use secondary streams on the worst cameras, as the res is terrible , 360 x 240 and nopt encountering issues with using main stream for detectionOperating systemHassOS Install methodHassOS Addon Network connectionMixed Camera make and model24 Mixed Manufacturers Any other information that may be helpfulHave had absolute blast discovering Home assistant and Frigate and have saved 20+ mixed quality Chinese IP cameras , that without Frigate where destined for landfill As some background to fine tuning I've done already, setting detect width and height to secondary FFPROBE width and height info , helped hugely, CPU was running at 80 to 85% , setting detect to sub streams , dropped CPU usage to ~70% , memory usage, currently seems ok at 50 to 60% utilisation, pre setting detect to sub streams was around 70% Is there any tweaking I can do further or any way to allow the GPU to assist here ? |
Beta Was this translation helpful? Give feedback.
Answered by
NickM-27
Jun 11, 2024
Replies: 2 comments 12 replies
-
|
easiest way to check this would be to see screenshots of the system page. You are restreaming both the high and low res streams which uses more CPU |
Beta Was this translation helpful? Give feedback.
12 replies
-
|
|
Beta Was this translation helpful? Give feedback.
-
|
|
Beta Was this translation helpful? Give feedback.
-
|
Appreciate any input to learn here Nick and compliments and huge praise to you and the Frigate crew !! Absolutely loving it :) |
Beta Was this translation helpful? Give feedback.
-
|
this looks pretty reasonable for CPU usage. The GPU can't run everything, the CPU still has to move the data to and from the GPU, convert the pixel format, etc. as these are things the GPU can't help with |
Beta Was this translation helpful? Give feedback.
Answer selected by
Bashewa
-
|
What does the GPU actually do ? Even in times of high feet traffic etc., GPU is max around 4% load and max I've seen the corals at is 25% |
Beta Was this translation helpful? Give feedback.
-
|
the GPU decodes the frames, meaning it takes the compressed frame data from the camera and exports individual frames.
it will not directly influence either, except a slight increase in CPU usage for tracking the new object types |
Beta Was this translation helpful? Give feedback.
-
|
@NickM-27 |
Beta Was this translation helpful? Give feedback.
-
|
To add on to my first question and current setup , I upgraded from the 4 core NUC to a 6 core HP tiny pc i5 8th gen , with only one coral USB ( second one didn’t seem to help allot ) and now at 26 cameras , processor load is roughly 50% , so to me , the extra 2 cores seemed to drop processor load nicely |
Beta Was this translation helpful? Give feedback.
-
|
In my case, I was lazy and thought that using only mainstream will do it all. I added lines and separated mainstream for recording and substream for detection and the CPU it lower. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
this looks pretty reasonable for CPU usage. The GPU can't run everything, the CPU still has to move the data to and from the GPU, convert the pixel format, etc. as these are things the GPU can't help with